def get_orders_restricted_projected(startdate, enddate):
"""Execute the equivalent of the following SQL query, querying the CSV file orders.csv:
SELECT
orderkey, custkey
FROM
orders
WHERE
startdate <= orders.orderdate and orders.orderdate < enddate
:param startdate: The lower boundary (inclusive) of the date range for the column 'orderdate'.
:param enddate: The upper boundary (exclusive) of the date range for the column 'orderdate'.
"""
timer = FunctionTimer()
short_prints = True
def read_date(date_str: str) -> datetime:
return datetime.strptime(date_str, "%Y-%m-%d").date()
orders_types = {"orderkey": int, "custkey": int, "orderdate": read_date, "totalprice": float, "shippriority": int}
orders = csv.import_csv("orders.csv", orders_types)
timer.lap("orders", short=short_prints)
def select_dates(rel) -> bool:
orderdate = rel("orderdate").value
return (startdate <= orderdate) and (orderdate < enddate)
orders_restricted = sets.restrict(orders, select_dates)
timer.lap("orders_restricted", short=short_prints)
orders_restricted_projected = clans.project(orders_restricted, "orderkey", "custkey")
timer.end("orders_restricted_projected", short=short_prints)
return orders_restricted_projected
def test_restrict(self):
self.assertIs(restrict(Undef(), Undef()), Undef())
s1 = Set(1, 2, 3)
self.assertRaises(TypeError, lambda: restrict(s1, Undef()))
self.assertRaises(TypeError, lambda: restrict(s1, 1))
self.assertRaises(TypeError, lambda: restrict(s1, 'a'))
self.assertEqual(restrict(s1, lambda x: x.value < 3), Set(1, 2))
self.assertEqual(restrict(s1, lambda x: x.value > 1), Set(2, 3))
def test_restrict(self):
self.assertIs(restrict(Undef(), Undef()), Undef())
s1 = Set(1, 2, 3)
self.assertRaises(TypeError, lambda: restrict(s1, Undef()))
self.assertRaises(TypeError, lambda: restrict(s1, 1))
self.assertRaises(TypeError, lambda: restrict(s1, 'a'))
self.assertEqual(restrict(s1, lambda x: x.value < 3), Set(1, 2))
self.assertEqual(restrict(s1, lambda x: x.value > 1), Set(2, 3))
def get_orders_restricted_projected(startdate, enddate):
"""Execute the equivalent of the following SQL query, querying the CSV file orders.csv:
SELECT
orderkey, custkey
FROM
orders
WHERE
startdate <= orders.orderdate and orders.orderdate < enddate
:param startdate: The lower boundary (inclusive) of the date range for the column 'orderdate'.
:param enddate: The upper boundary (exclusive) of the date range for the column 'orderdate'.
"""
timer = FunctionTimer()
short_prints = True
def read_date(date_str: str) -> datetime:
return datetime.strptime(date_str, '%Y-%m-%d').date()
orders_types = {
'orderkey': int,
'custkey': int,
'orderdate': read_date,
'totalprice': float,
'shippriority': int
}
orders = csv.import_csv('orders.csv', orders_types)
timer.lap('orders', short=short_prints)
def select_dates(rel) -> bool:
orderdate = rel('orderdate').value
return (startdate <= orderdate) and (orderdate < enddate)
orders_restricted = sets.restrict(orders, select_dates)
timer.lap('orders_restricted', short=short_prints)
orders_restricted_projected = clans.project(orders_restricted, 'orderkey',
'custkey')
timer.end('orders_restricted_projected', short=short_prints)
return orders_restricted_projected